Python mxnet.ndarray.NDArray() Examples

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name]) 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def get_outputs(self, merge_multi_context=True):
        """Get outputs of the previous forward computation.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        outputs = [[exec_.outputs[i] for exec_ in self.execs]
                   for i in range(len(self.execs[0].outputs))]
        if merge_multi_context:
            outputs = _merge_multi_context(outputs, self.output_layouts)
        return outputs 

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name]) 

def __init__(self, interval=1, pattern='.*', sort=False):
        def stat(x):
            return x.shape
        self.stat_func = stat
        self.interval = interval
        self.activated = False
        self.queue = []
        self.step = 0
        self.exes = []
        self.re_prog = re.compile(pattern)
        self.sort = sort
        def stat_helper(name, array):
            array = ctypes.cast(array, NDArrayHandle)
            array = NDArray(array, writable=False)
            if not self.activated or not self.re_prog.match(py_str(name)):
                return
            self.queue.append((self.step, py_str(name), stat(array)))
        self.stat_helper = stat_helper 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_outputs(self, merge_multi_context=True):
        """Get outputs of the previous forward computation.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        outputs = [[exec_.outputs[i] for exec_ in self.execs]
                   for i in range(len(self.execs[0].outputs))]
        if merge_multi_context:
            outputs = _merge_multi_context(outputs, self.output_layouts)
        return outputs 

def get_outputs(self, merge_multi_context=True):
        """Get outputs of the previous forward computation.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        outputs = [[exec_.outputs[i] for exec_ in self.execs]
                   for i in range(len(self.execs[0].outputs))]
        if merge_multi_context:
            outputs = _merge_multi_context(outputs, self.output_layouts)
        return outputs 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name]) 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def get_input_grads(self, merge_multi_context=True):
        """Get the gradients with respect to the inputs of the module.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[grad1, grad2]`. Otherwise, it
        is like `[[grad1_dev1, grad1_dev2], [grad2_dev1, grad2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert self.inputs_need_grad
        if merge_multi_context:
            return _merge_multi_context(self.input_grad_arrays, self.data_layouts)
        return self.input_grad_arrays 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_input_grads(self, merge_multi_context=True):
        """Get the gradients with respect to the inputs of the module.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[grad1, grad2]`. Otherwise, it
        is like `[[grad1_dev1, grad1_dev2], [grad2_dev1, grad2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert self.inputs_need_grad
        if merge_multi_context:
            return _merge_multi_context(self.input_grad_arrays, self.data_layouts)
        return self.input_grad_arrays 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name]) 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def get_outputs(self, merge_multi_context=True):
        """Get outputs of the previous forward computation.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        outputs = [[exec_.outputs[i] for exec_ in self.execs]
                   for i in range(len(self.execs[0].outputs))]
        if merge_multi_context:
            outputs = _merge_multi_context(outputs, self.output_layouts)
        return outputs 

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name]) 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_input_grads(self, merge_multi_context=True):
        """Get the gradients with respect to the inputs of the module.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[grad1, grad2]`. Otherwise, it
        is like `[[grad1_dev1, grad1_dev2], [grad2_dev1, grad2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert self.inputs_need_grad
        if merge_multi_context:
            return _merge_multi_context(self.input_grad_arrays, self.data_layouts)
        return self.input_grad_arrays 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name]) 

def backward(self, out_grads=None):
        """Run backward on all devices. A backward should be called after
        a call to the forward function. Backward cannot be called unless
        `self.for_training` is `True`.

        Parameters
        ----------
        out_grads : NDArray or list of NDArray, optional
            Gradient on the outputs to be propagated back.
            This parameter is only needed when bind is called
            on outputs that are not a loss function.
        """
        assert self.for_training, 're-bind with for_training=True to run backward'
        if out_grads is None:
            out_grads = []

        # for i, (exec_, islice) in enumerate(zip(self.execs, self.slices)):
        for i, exec_ in enumerate(self.execs):
            out_grads_slice = []
            exec_.backward(out_grads=out_grads_slice) 

def get_input_grads(self, merge_multi_context=True):
        """Get the gradients with respect to the inputs of the module.

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the outputs
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[grad1, grad2]`. Otherwise, it
        is like `[[grad1_dev1, grad1_dev2], [grad2_dev1, grad2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert self.inputs_need_grad
        if merge_multi_context:
            return _merge_multi_context(self.input_grad_arrays, self.data_layouts)
        return self.input_grad_arrays 

def get_states(self, merge_multi_context=True):
        """Get states from all devices

        Parameters
        ----------
        merge_multi_context : bool
            Default is `True`. In the case when data-parallelism is used, the states
            will be collected from multiple devices. A `True` value indicate that we
            should merge the collected results so that they look like from a single
            executor.

        Returns
        -------
        If `merge_multi_context` is `True`, it is like `[out1, out2]`. Otherwise, it
        is like `[[out1_dev1, out1_dev2], [out2_dev1, out2_dev2]]`. All the output
        elements are `NDArray`.
        """
        assert not merge_multi_context, \
            "merge_multi_context=True is not supported for get_states yet."
        return self.state_arrays 

def get_params(self, arg_params, aux_params):
        """ Copy data from each executor to `arg_params` and `aux_params`.

        Parameters
        ----------
        arg_params : list of NDArray
            target parameter arrays
        aux_params : list of NDArray
            target aux arrays

        Notes
        -----
        - This function will inplace update the NDArrays in arg_params and aux_params.
        """
        for name, block in zip(self.param_names, self.param_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(arg_params[name].dtype).copyto(arg_params[name])
        for name, block in zip(self.aux_names, self.aux_arrays):
            weight = sum(w.copyto(ctx.cpu()) for w in block) / len(block)
            weight.astype(aux_params[name].dtype).copyto(aux_params[name])